Abstract
In this paper, we focus on sparse signal recovery methods for data assimilation in groundwater models. The objective of this work is to exploit the commonly understood spatial sparsity in hydrodynamic models and thereby reduce the number of measurements to image a dynamic groundwater profile. To achieve this we employ a Bayesian compressive sensing framework that lets us adaptively select the next measurement to reduce the estimation error. An extension to the Bayesian compressive sensing framework is also proposed which incorporates the additional model information to estimate system states from even lesser measurements. Instead of using cumulative imaging-like measurements, such as those used in standard compressive sensing, we use sparse binary matrices. This choice of measurements can be interpreted as randomly sampling only a small subset of dug wells at each time step, instead of sampling the entire grid. Therefore, this framework offers groundwater surveyors a significant reduction in surveying effort without compromising the quality of the survey. © 2013 IEEE.
| Original language | English (US) |
|---|---|
| Title of host publication | 2013 10th IEEE INTERNATIONAL CONFERENCE ON NETWORKING, SENSING AND CONTROL (ICNSC) |
| Publisher | Institute of Electrical and Electronics Engineers (IEEE) |
| Pages | 176-182 |
| Number of pages | 7 |
| ISBN (Print) | 9781467352000 |
| DOIs | |
| State | Published - Apr 2013 |
| Externally published | Yes |
Bibliographical note
KAUST Repository Item: Exported on 2020-10-01Acknowledgements: The authors would like to thank Dr. Ibrahim Hoteit andMr Mohamad El Gharamti at KAUST for providing us withthe code to simulate 2D contaminant flow in the groundwater.We also thank Hasan Arshad Nasir who helped us greatly inthe initial phase of the project. This work was carried outat the Laboratory of Cyber Physical Networks and Systems(CYPHYNETS) at LUMS under a project funded by theEnvironmental Protection Agency (EPA) of the Governmentof Punjab, Pakistan.
This publication acknowledges KAUST support, but has no KAUST affiliated authors.